When driving a transformer, you don't want the core flux to saturate. This happens when the magnetic field can no longer increase in response to the applied voltage. This is characterized as the Voltage-Time product, or V-t (or sometimes ET). To calculate the required V-t product for a transformer you multiply the maximum voltage applied by the time the voltage is applied. Since the voltage is applied in a given direction for half the cycle time, the formula becomes:
Vtmin = Vmax / (2 * Fmin)What happens if you exceed the transformer's V-t product and the core saturates? Since the magnetic flux stops increasing with increasing primary winding current there is no increase in current in the transformer's secondary winding. Aside from the loss of efficiency, the increase in current turns to heat. Too much heat and the magic smoke escapes from the transformer (we all know electronics operate on magic smoke, because when the magic smoke escapes the electronics stop working).
The SN6505 I'm going to use to drive the VFD filament transformer has two variants: the SN6505A oscillates at 160 KHz, while the SN6505B oscillates at 424 kHz. The Canon blocking oscillator ran at about 133 KHz. Noritake-Itron recommends the filament be driven "with a frequency between 10kHz and 200kHz". It seemed to me that 424 kHz was kinda fast, so I thought I'd try the lower-speed variant of the SN6505. But there's a problem.
I'm driving the transformer with a nominal 3.3 volts, but assuming a 10% tolerance it could be as low as 2.97 or as high as 3.63 volts. The SN6505 implements spread-spectrum clocking so the actual frequency will vary from 138 to 203 kHz for the A variant, and from 363 to 517 kHz for the B variant. Since the transformer's V-t product needs to be higher than the applied signal we care most about the highest voltage and lowest frequency.
The transformer I'm using, the Halo TGM-210NSLF, is rated at 11 V/µs. It was designed for use with the MAX253 transformer driver which oscillates at a minimum of 250 kHz in its high frequency mode. With a 3.63 volt supply the minimum V-t required is 7.26 V/µs. This is the configuration I've been using in my tests.
Driven with 3.63 volts at a minimum frequency of 363 kHz the SN6505B requires a transformer with a minimum V-t product of just 5 V/µs. I tested this briefly and found it worked just fine. The SN6505A has a minimum frequency of just 138 kHz, however, which would require a transformer with a V-t product of at least 13 V/µs. That's too high for the 11 V/µs-rated Halo transformer and is a magic smoke containment issue.
I kinda doubt DigiKey would accept a return of a cut tape of 3 chips, even in the unopened static bag. Maybe I'll put 'em on eBay. Anyone want to buy a few SN6505A chips?
I guessI'm going to have to cannibalize my current test circuits, not only for the Halo transformer, but for the SN6505B driver as well. Good thing I fixed my hot air soldering station!
Fortunately, Mouser stocks the SN6505B, and I've added a couple to my pending order.